Details in this account are generalised and figures are omitted where they would identify the client.
The situation
An installation and maintenance business working across the region for commercial customers. Quotes were produced by the owner and one senior estimator, from a spreadsheet, taking around forty minutes each.
The consequences were predictable. Quotes went out days after the enquiry. During busy periods the backlog grew. Nobody else could produce one, so holidays created a bottleneck. And because each quote was composed rather than assembled, prices varied between the two of them for the same work.
The trigger was losing a piece of work to a competitor who had quoted within a day.
Extracting the pricing rules
This was the project. The software was the reason to do it.
We took two days with the owner and the estimator, going through around sixty past quotes and asking, for each one, why it was priced as it was.
What emerged: a base rate per item, adjusted by quantity, by access difficulty, by distance from the depot, by whether the work was out of hours, and by a customer-specific arrangement for four long-standing accounts.
Most of these were not written down anywhere. Two of them the estimator applied without being able to state the rule until we found three examples and worked backwards.
Several inconsistencies surfaced. The two of them applied the distance adjustment differently, and had done for years. Neither was wrong exactly; they had simply never compared.
The pricing conversation was worth more than the software. Two people had been quoting differently for years and neither knew, because nobody had ever put their quotes side by side.
What we built
A priced catalogue of items and standard packages, with the adjustment rules encoded.
A quote is assembled by selecting items and quantities and answering a small number of questions about the job — access, distance, timing. The price is calculated, the document is generated, and it goes out as a link the customer can accept with a button.
Sending a quote creates a follow-up action with a date, automatically.
Why the first version was rejected
The initial build calculated the price and did not permit changing it. The reasoning had been that consistency was the point.
The estimator refused to use it, and was right to.
Experienced estimators make judgements the rules do not capture — a difficult customer, a job that will lead to more, a site where something about the description does not add up. Removing that discretion did not make pricing more consistent; it made the system unusable by the person who knew most.
What replaced it: the price is calculated, and an authorised user can override it. The override is recorded with a reason and appears in a monthly report.
That change did two things. It made the system acceptable to the people who had to use it, and it made discounting visible for the first time — because an override against a calculated price is a recorded event, where a low number typed into a blank spreadsheet is not.
What happened
Quotes now take a few minutes and can be produced by any of four people rather than two.
The follow-up prompts turned out to matter as much as the speed. Quotes had previously been chased when somebody remembered, which meant a proportion never were. Every quote now has a dated action against it.
The override report produced an uncomfortable and useful finding in its first quarter: a particular category of work was being discounted routinely by both estimators, which suggested the base price for it was wrong. It was adjusted.
What we would do differently
Build in the override from the start. We designed for consistency and had to be corrected by the people doing the job. A morning watching them quote would have shown us.
Release the catalogue before the system. Simply having a written, agreed price list would have delivered value on its own, weeks earlier.
Connect to accounting sooner. Accepted quotes still needed manual entry into the accounts for the first few months, which was an obvious gap we had scheduled for later.
The general lessons
Where a system meets professional judgement, calculate and allow override with a recorded reason. Do not remove the judgement.
The rules extraction is the valuable work. Expect it to surface inconsistencies, and treat those as findings rather than problems.
Attaching follow-up to sending is frequently worth more than the time saved producing the document.
Our automation team builds quoting systems for UK trades businesses — see also quoting without the retyping. Start a conversation.